Anxiety and mood disorders are highly prevalent, often co-morbid, and when so, are associated with increased morbidity and disability. In particular, anxiety in the context of depression predicts worse outcome with a range of treatments. The commonality of excessive, uncontrollable negative emotion across anxiety and mood disorders suggests a core deficit in emotional reactivity and regulation. We, and others, have delineated a neurobehavioral system, involving the anterior cingulate, amygdala, and lateral prefrontal cortex, that is involved in emotional reactivity and regulation. Both activity ad connectivity within this circuit are perturbed in patients with anxiety and mood disorders, supporting the hypothesis that dysfunction in the handling of emotion lies at the core of these disorders, and that validated tools are now available for objectively assessing this neural system. These findings constitute a robust scientific foundation for novel intervention approaches aimed at improving emotion regulation, which could be delivered over the internet, thereby also increasing availability of effective treatments. We propose to pilot a novel, internet delivered neuroplasticity-based neurobehavioral intervention, which improves emotion regulation by 1) instilling a bias towards positive stimuli, 2) improving resistance to emotional distraction, and 3) enhancing executive functioning more generally (e.g. working memory, task switching, resisting interference), as these are required for successful implementation of emotion regulation (ER). In the R21 phase, thirty medication-free patients with anxious depression (total Ham-D17e16, Ham-D17 anxiety subscale 7) will be given 60 days of internet-delivered training. The aim of this phase is to optimize the training and assessment procedures (symptom, behavioral and fMRI), as well as to provide initial evidence for generalization of the training to measures relevant for anxiety and depression. In the R33 phase, sixty medication-free patients with anxious depression will be randomized to receive the 60-day ER neurobehavioral intervention (with adjustments if indicated by the R21) or an active control arm of online tasks that are similarly engaging but provide no emotion regulatory benefits. The aim of this phase is to provide initial evidence supporting the utility of an optimizd version of the neurobehavioral intervention, focusing on effect size estimation and delineation of recruitment and implementation parameters for a large-scale clinical trial. We foresee our internet-delivered, adaptive neurobehavioral training for ER as being applicable for a range of psychiatric disorders, and in particular for mood or anxiety disorders. Successful completion of the proposed study would provide compelling initial evidence for the utility of this novel intervention approach, guided by a neuroscientific understanding of core deficits in anxiety and depression, for a larger and more definitive clinical trial.